In 2008, Measurement Computing released a self-contained datalogger that will record temperatures from a thermocouple continuously.
The USB-501TC will read 32,510 temperature samples, each time-stamped. It will read K, J, and T-type thermocouples, and ships with the K-type thermocouple shown in the picture. The temperature readings can be taken at a number of intervals ranging from 1 second out to 12 hours.
Under the cap you find a standard USB plug, and the housing can be opened to access the replaceable 3.6V lithium battery (1/2 AA size). The battery isn’t something you’ll find at the local drug store, but replacements can be ordered from a number of online vendors.
The quoted resolution of the datalogger is 0.5°C (though they actually report in 0.1°C increments) and the accuracy is supposed to be +/- 1°C. It’s up to you to decide if you can live with that sort of accuracy, but the stated specs seem to hold up in practice. I calibrated 5 dataloggers against a NIST-traceable 0.1°C-resolution alcohol thermometer and got the following results:
The data shown here are thermometer temperature and the average of 5 USB-501TC thermocouple temperatures measured at the same time points, with both submerged in a well-stirred water bath. Over the range of biologically-relevant temperatures, the USB-501TC fitted with a 36-gauge T-type copper-constantan thermocouple reads within 1-2°C of the calibrated thermometer. It is important to note that I measured these temperatures while slowly heating the water bath (1°C per minute), so some of the error present is due to response-time differences between the thermometer and the thermocouple. The actual correspondence between the two is likely slightly better than what I show here, but you’re still going to be dealing with a potential error of +/- 1°C. The error bars represent +/- 1 standard deviation.
In the above graph I’ve also plotted a line of unity (ideal 1-to-1 correspondence between the x and y axis measurements) and that the USB-501TC data deviate slightly from the ideal 1-to-1 slope (actual slope = 1.0633). At temperatures below ~40°C, the USB-501TC’s tended to read slightly low, and above 40°C they read slightly high. This may be better illustrated with a graph showing the difference between the thermometer and thermocouple versus thermometer temperature:
When the values are negative in this figure, the thermocouple datalogger is reading a lower temperature than the thermometer. As you go from 0°C to 40°C, that difference shrinks. Again, each value is the average of 5 USB-501TC’s and error bars are 1 standard deviation.
The USB-501TC does what it claims to do. But why would I care? There are plenty of self-contained temperature dataloggers out there (iButtons, TidBits) that do the same thing (measure temperature), and are more water resistant than this unit. But there are often times when you simply can’t fit an iButton or Tidbit in whatever it is you’re trying to measure, whereas you can usually shove the tip of a thermocouple into anything. The beauty of the USB-501TC is that it’s appropriately sized so that it can be shoved in a water-tight housing, allowing it to fill roughly the same role as those other dataloggers, but in a package with a replaceable battery, substantial memory space, and that tiny thermocouple tip, all for roughly $90. It should be noted that the software for downloading and launching the dataloggers only works on Windows 2000/XP/Vista (no Mac version, and it doesn’t work on 64-bit Vista/7 as of early 2010).
The diameter of the datalogger is just small enough to fit inside schedule-40 PVC plumbing pipe (but not schedule 80). I made the housings below to hold these dataloggers in the field.The housing consists of a length of 1″ pipe, a female threaded x slip connector, a 1″ pipe plug, and a 1″ round pipe cap. In the round pipe cap, I drilled and tapped a 3/8″ NPT thread to fit the liquid-tight cord grip (McMaster-Carr PN# 69915K52), through which the thermocouple wire exists the housing. The entire housing is roughly 7.5″ long (I measured with my gringo ruler).
There are a few things to watch with this housing. The main issue is where the thermocouple wire passes into the housing. I waterproofed this connection by covering the 36-gauge thermocouple wire with several layers of adhesive-lined shrink tubing to build up the diameter of the wire to the point where the cord-grip could actually grab and seal the wire. By doing this, you limit the adjustability of the length of wire sticking out of the housing, unless you build up heat shrink tubing at several spots. You could do this in the field with a cigarette lighter if needed. The secondary issue is that you must waterproof the end of the thermocouple wire so that water can’t wick up the thermocouple jacketing and get into the housing.Finally, you need some reliable way to strap the housing down in the field. We have used galvanized perforated strap thus far for short-term deployments.
Finally, there is also a version of the datalogger with a LCD readout, the USB-501TC-LCD, so that you can read the current temperature (or max/min temperature) while the datalogger is busy working. Additionally, there is a whole family of dataloggers based off the same basic architecture, the USB-500 series, including units that measure temperature without an external probe, higher-resolution loggers, and combination relative humidity/temperature loggers, all between $50 and $150.